The FCA looks at agentic finance and asks who regulates the agent
A landmark UK review says consumers will increasingly delegate money decisions to AI, and wants an answer within six months on whether the models themselves need regulating.
- The UK Financial Conduct Authority published a landmark review of AI in retail finance on July 6, led by executive director Sheldon Mills.
- It asks the government to examine whether large language models, naming ChatGPT, Claude and Gemini, should fall under FCA rules.
- The review anticipates consumers delegating financial decisions to agents acting within customer-set parameters, and recommends assessing the perimeter question within three to six months.
- Mills called the dynamic "an arms race" in an interview with the Financial Times.
Britain's financial regulator has looked at where AI in finance is heading and concluded the destination is agents, not chatbots. The Financial Conduct Authority's review of AI in retail financial services, published July 6 and led by executive director Sheldon Mills, asks the UK government to examine whether general-purpose models such as ChatGPT, Claude and Gemini should be brought inside the FCA's regulatory perimeter.
Delegation is the operating assumption
The review's core forecast is the one this publication tracks every week: AI moving past recommendations into agents that autonomously manage money within parameters the customer sets. "AI will reshape consumer financial journeys, with people increasingly delegating to AI applications that act on their behalf," Mills writes in the review. He was blunter with the Financial Times: "It is an arms race."
The regulatory tripwire is subtle. As Mills put it, "continuous and adaptive recommendations may start to look like regulated advice." A model that keeps refining its guidance as it watches your accounts is doing something UK law already has a category for. It just is not held to that category yet.
A six-month clock
Per Reuters, Mills recommends the FCA assess within three to six months whether its boundaries need redrawing, and the review flags a concentration risk regulators keep circling: the global financial system increasingly leaning on a handful of model providers. The Bank of England has separately signaled that autonomous agent systems may need bespoke rules, per the same Reuters reporting.
Why GaaS vendors should read this one
Until now, financial regulation has mostly landed on the banks and brokers that deploy AI, not the companies that supply it. A G7 regulator openly asking for jurisdiction over the model layer is a different posture. If it lands, every vendor selling agentic outcomes into UK finance inherits a compliance surface that looks less like a software vendor's and more like a financial firm's. It also lands the same week China's agent-specific rules took force, which makes two major economies moving on agents in one news cycle.
Sources: PYMNTS, Reuters, BanklessTimes.